A three-state model for the Photo-Fries rearrangement Josene Toldo, Mario Barbatti, Paulo F. B. Gonçalves

To cite this version:

Josene Toldo, Mario Barbatti, Paulo F. B. Gonçalves. A three-state model for the Photo-Fries rearrangement. Physical Chemistry Chemical Physics, Royal Society of Chemistry, 2017, 29 (19), pp.19103-19108. ￿10.1039/C7CP03777E￿. ￿hal-02288764￿

HAL Id: hal-02288764 https://hal-amu.archives-ouvertes.fr/hal-02288764 Submitted on 16 Sep 2019

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés. Please do not adjust margins

A three-state model for the Photo-Fries rearrangement

Josene M. Toldo,a,b Mario Barbattib and Paulo F. B. Gonçalvesa

A three-state model for the Photo-Fries rearrangement (PFR) is proposed based on multiconfigurational calculations. It provides a comprehensive mechanistic picture of all steps of the reaction, from the photoabsorption to the final tautomerization. The three states participating in the PFR are an aromatic 1ππ*, which absorbs the radiation; a pre- dissociative 1nπ*, which transfers the energy to the dissociative region; and a 1πσ*, along which dissociation occurs. The transfer from 1ππ* to 1nπ* involves pyramidalization of the carbonyl carbon, while transfer from 1nπ* to 1πσ* takes place through CO stretching. Different products are available after a conical intersection with the ground state. Among them, a recombined radical intermediate, which can yield ortho-PFR products after an intramolecular 1,3-H tunneling. The three- state model is developed for phenyl acetate, the basic prototype for PFR, and it reconciles theory with a series of observations from time-resolved spectroscopy. It also delivers a rational way to optimize PFR yields, since, as shown for four different systems, diverse substituents can change the energetic order of the 1ππ* and 1nπ* states, preventing or enhancing PFR.

phenol. In addition, the reaction quantum yield of rearranged Introduction products is strongly influenced by solvent polarity as well as by the presence of electron donor or acceptor in the aromatic moiety.19, 27, 32, 44, 45 Photo-Fries rearrangement (PFR)—a photochemical conversion of aryl esters to ortho- and para-hydroxyphenones (Scheme 1)—is a key step in the synthesis of a large number of compounds.1-4 It also plays an important role in the design of functional polymers5-8 and in the photodegradation of drugs9, 10 and agrochemicals.11-13 Compared to its thermal version, the Lewis-acid catalyzed Fries rearrangement, PFR has an additional benefit of being a greener synthetic route, since it can be achieved under milder conditions.3, 14, 15 Given its importance for synthesis, it is not surprising that PFR has been the subject of numerous investigations in the past.15-36 Nevertheless, the conceptual theoretical knowledge of this reaction is still incipient36- 38 and, as we shall see, even the full set of electronic states involved in the reaction has not been yet identified. Experimental observations have established that PFR takes place 36, 37 in the lowest singlet state (S1) although, in some cases, a contribution from upper triplet states is also expected.34, 39-41 The homolytic cleavage of the OC–O bond gives rise to a carbonyl and phenoxyl radical pair. The subsequent recombination leads to the starting ester and to cyclohexadienone intermediate, which Scheme 1. General scheme of Photo-Fries rearrangement. For phenyl tautomerizes to yield the rearranged products. The final step is a acetate, R = methyl. hydrogen shift, which can proceed either via tunneling or through Further insights into the early events in PFR of phenyl acetate solvent rearrangement.42, 43 Alternatively, the radicals can escape (PA) in cyclohexane has been provided using transient electronic and from the solvent cage leading to formation of the corresponding vibrational absorption spectroscopies. Pumping at 267 nm, they show radical pairs being formed within 28 ps, although phenoxyl a. Department of Physical Chemistry, Federal University of Rio Grande do Sul, Av. radicals are observed as early as 15 ps.36 Two-color femtosecond Bento Gonçalves, 9500, Porto Alegre-RS, CEP 91501-970, Brazil. pump-probe spectroscopy pumped at 258 nm, revealed that the S1 b. Aix Marseille Univ, CNRS, ICR, Marseille, France. 1 *Corresponding authors : [email protected] (JMT), mario.barbatti@univ- state of para-tBu-PA, also in cyclohexane, is depopulated via * amu.fr (MB), [email protected] (PFBG). within just 2 ps and the dissociated radicals recombine within 13 ps.37 Electronic Supplementary Information (ESI) available: Molecular orbitals, absorption spectra, reaction path, energies, and Cartesian coordinates. See In contrast to a large number of experimental studies, the last DOI: 10.1039/x0xx00000x theoretical investigation on PFR was delivered by Grimme, in 1992,

Please do not adjust margins Please do not adjust margins

using semi-empirical methods.38 In that work, barriers between 0.9 was used and an imaginary level shift48 of 0.1 a.u was applied to deal and 1.2 eV were found for PA photodissociation starting from a 1nπ with intruder states. The ANO-S-VDZP49 basis set was employed in all

* state. Such large barriers are clearly incompatible with the calculations. The S1/S0 conical intersection was initially optimized at measured picosecond time scale of the process.36, 37 Moreover, still the CASSCF level. Due to the usual energy split when CASPT2 is due to methodological limitations of that early work,38 the relative computed for such geometries,50 the intersection was further importance of dissociation along 1nπ* versus 1πσ* could not be relaxed at CASPT2 level. Thus, starting from the CASSCF geometry, clearly stated. In fact, the lack of high-level theoretical information restricted optimizations along the CO–O bond (R) were done at the on PFR is such that even the character of the initial excited state— MS3-CASPT2(6,6) level, until the S1 and S0 states became 1nπ* or 1ππ*—has still been under debate.36-38 degenerated. For this final intersection geometry, energies were Given the knowledge gap between theory38 and the most recent computed at MS3-CASPT2(14,12). The subsequent pathway after the experimental works,36, 37 our aim has been to provide a CI was optimized in the ground state at the CASSCF level, still along comprehensive picture of PFR, based on high-level constrained values of R. The branch yielding the radical pair was multiconfigurational theoretical methods, applied to PA in the calculated starting from large values of R, while the branch giving rise gas phase, the minimum prototype to understand PFR. to PFR was calculated systematically increasing R starting from the CI The multiconfigurational theoretical approach has allowed us to structure. All calculations were carried out using MOLCAS 8 clarify several the following questions: Which state is initially program.51 populated? How is the energy transferred from the Franck-Condon region to the dissociative pathway? What are the electronic states involved and their multiplicities during dissociation? Is there any Results and discussions relevant conical intersection along the way? Why does the The PFR mechanism photoexcited population branches into dissociated and The analysis of relaxed reaction pathways in the excited states rearranged species? How does tautomerization occur? computed with MS-CASPT2//CASSCF shows that after Our results revealed that PFR involves three electronic excited photoexcitation, PFR takes place through the S state involving three states arranged along a specific topography that allows transferring 1 diabatic characters. A schematic potential energy profile the photoenergy from the aromatic to the carbonyl region. To summarizing this three-state model is shown in Figure 1 for PA. Along further explore this three-state model for PFR, we extended the the solid lines, the OC-O bond distance is the main reaction calculations for three other aromatic esters containing an coordinate, while along the dashed curve, the hydrogen shift amino group instead of a methyl group attached to carbonyl between the oxygen and ortho carbon is the main reaction moiety. As result, we succeed in providing a solid conceptual basis coordinate. Although the relative energies in this figure correspond for a class of reactions important for organic, polymer, and to those for PA, this three-sates profile is still valid for other environmental chemistry, answering questions that have hindered molecules undergoing PFR, as discussed later. progress in these fields and laying the groundwork for interpreting four decades of experiments.

Computational details Theoretical calculations were carried out using MS-CASPT2//CASSCF protocol,46 in which energies are computed at the multi-state complete active space second-order perturbation theory (MS-CASPT2) on structures optimized at the complete active space self-consistent field level (CASSCF). Critical points (minima, transition states, and conical intersection) and reaction paths were optimized with an active space including 14 electrons in 12 orbitals and state-averaged over three states (SA3-CASSCF(14,12)). Cartesian coordinates for all these structures are given in the Supporting Information (SI). The active space for PA was composed of seven occupied and five virtual orbitals: 4π and 4 π *, 1 orbital pair σ/σ* along the OC–O bond, and two non-bonded Figure 1. Schematic overview of the three-state model for PFR electrons pairs, one in the oxygen of the carbonyl group and applied to PA. The insets show the main orbital transitions of the another in the oxygen bonded to the phenyl ring (see SI1). This states involved in the PFR. Along the solid lines, the OC-O bond active space was kept during the subsequent geometry distance is the main reaction coordinate. Along the dashed curve, the optimizations. For the remaining molecules investigated—phenyl hydrogen shift between the oxygen and ortho carbon is the main carbamate, metoxyphenyl carbamate, and 2-isopropoxyphenyl N- reaction coordinate. methylcarbamate—the active space included the same set of orbitals as described for PA. For the description of the excited states, the CASSCF was still averaged over three states, whereas the energy was corrected with MS-CASPT2 over 7 states. In the CASPT2 calculations, the standard IPEA parameter47 of 0.25 a.u.

Please do not adjust margins Please do not adjust margins

1 Table 1. Vertical excitation energy (ΔEvert) from the ground state * excited state is exceedingly large, about 0.8 eV. On the other minimum, oscillator strength (fosc), and main configuration of the hand, no barrier was found when the photodissociation started from lowest excited states calculated using MS-CASPT2//CASSCF. the pyramidalized minimum in the 1n* state. As result, pyramidalization of the acetyl moiety is required for State ΔEvert (eV) fosc Configuration photodissociation and, in this way, the 1n* state can be considered  as a pre-dissociative state. S1 4.82 0.0028  The small barrier between 1* and 1n* (<0.38 eV) and the S 5.86 0.1746  2 barrierless transition from 1n* to 1* is the key to understanding  S3 6.03 0.0020 n how the 1* can be populated in just 2 ps, as experimentally observed.37 We mentioned above that dissociation barriers between 1 38 In the Franck-Condon region, the S1 state has a dark * 0.9 and 1.2 eV were predicted for PA in an early theoretical work. character (Table 1; see SI2 too). A second 1* state with larger Such extremely large barriers arose from the geometric constraints oscillator strength appears with energy close to the 1n* state, but imposed in that study, which carried out a limited exploration of the about 1 eV higher than S1. Thus, pumping PA at 267 nm (4.64 eV), as dihedral angle and did not consider pyramidalization at all. 36 1 1 done in Ref. , excites the dark * state. Note yet that while the A conical intersection between the dissociative πσ* and the S0 excited electronic density of the 1* states are located on the surfaces is indicated in Figure 3 close to 2.2 Å. This crossing is a aromatic ring, that of the 1n* is mainly at the acetyl moiety (see common feature for PA38 and related molecules, as phenol52 and molecular orbitals in Figure 1). para-tBut-phenyl acetate.37 After the intersection, the reaction path

During the optimization of S1, two minima were found (see Figure splits in three ways, with one branch returning to the S0 parent 1 2). The first one (S1-PL) has a Cs geometry with a planar conformation molecule, another branch following the πσ* state originating a of the acetyl moiety. It features a 1* character still located on the cyclohexanone-acetyl radical pair, and a third branch forming a aromatic ring. The second minimum (S1-PYR) has a significant stable cyclohexadienone intermediate. pyramidalization of the carbon atom on the acetyl moiety, displacing The population of these three branches is the key step for the the oxygen out of the molecular plane. It features a 1n* character PFR yield and it should depend on the particular excited-state located in the acetyl moiety only. The S1-PYR minimum is 0.12 eV topography for each molecule and on the solvent as well. If the below the S1-PL minimum. A linear interpolation in internal conical intersection is reached with a high excess of kinetic energy, coordinates shows that the barrier to converting between them is radical-pair formation will dominate. But even in this case, PFR can smaller than 0.38 eV (shown in SI3). still take place as the solvent cage53 may inhibit dissociation and induce recombination of the radical into substituted 2,4- and 2,5- cyclohexadienone. In cyclohexane, the formation of cyclohexanone intermediate is found to occur between 13 ps37 (para-tBut-PA excited at 258 nm) and 42 ps36 (PA, 267 nm). In this latter case, the fraction of radical pairs formed that are expected to escape from the solvent cage is about 26% and most of the recombination products 36 (54%) will be S0 parent molecules.

Figure 2. MS-CASPT2 energies long the Photo-Fries rearrangement of PA. CI: Conical intersection; TS: Transition state.

According to previous works, photodissociation in PFR is mediated by a higher electronically excited state with 1σπ* character.36-38 It can be seen in Figure 3 that when the acetyl moiety is shifted along the reaction coordinate, this state stabilizes, Figure 3. MS-CASPT2 relaxed energy profile starting from the till becoming S0. Due to the uncertainty pointed out in previous 1 pyramidalized ( n*) S1 minimum (solid lines). The S1 energy profile works about the assignment of the electronic configuration of the 1 36 starting from the planar ( *) S1 minimum is also shown with a S1, both 1ππ 1 π * and n * states were investigated as starting point dashed line. The shaded curve indicates the 1* state. for the OC-O bond breaking. The S1-relaxed potential energy profiles in Figure 3 show that the dissociation is only possible after accessing the 1nπ* state because the barrier to reach the dissociative state starting from

Please do not adjust margins Please do not adjust margins

The last step of PFR is enolization of the cyclohexadienone eV in b and by 0.4 eV in c and d. We can rationalize this effect based intermediate to form the final substituted hydroxyphenone product. on the resonance structures that characterize the amino-substituted In the gas phase and in nonpolar aprotic solvents, H-shift is molecules, stabilizing the lone pairs and increasing the energies of intramolecular. As shown in Figure 2, the energy barrier for 1,3-H n* state. We note, however, that the correct description of the shift in the gas phase is 2.1 eV and can only be crossed via 1*-1n* energy gap requires a proper account of electron dynamic tunneling.36, 42 In methylcyclohexane at 293 K, for instance, the correlation, as revealed by the strong differences between CASSCF measured tunneling rate for the 1,3-H shift is 3.6 s-1.42 In protic and MS-CASPT2 results (SI4). solvents, as alcohols, the 1,3-H and 1,5-H shifts should be much faster, as they are aided by intermolecular interactions. 42

Table 1. Energy differences between the lowest singlet state and the lowest triplet states calculated at the S1-PL and S1-PYR minima using MS-CASPT2//CASSCF. The main configuration of the triplet states is shown as well.

State ΔE (eV) Config. ΔE (eV) Config. [a] [b] S1-PL-Tx S1-PYR-Tx 3 3 T1 -0.71 (*) -0.19 (n*)CO 3 3 T2 -0.09 (*) 0.44 (*) 3 3 T3 -0.02 (*) 1.71 (*)

[a] Relative to the S1-PL optimized geometry. [b] Relative to the S1- PYR optimized geometry.

Triplet states around the 1ππ* and 1nπ* minima were Figure 4. Energies of the planar 1(*) and pyramidal 1(n*) S also calculated to ascertain the spin multiplicity of the 1 minima of (a) phenyl acetate, (b) phenyl carbamate, (c) ortho- photodissociation process. Table 2 shows the energy of the triplet methoxyphenyl carbamate, and (d) 2-isopropoxyphenyl states compared to the energy of the lowest singlet state in S1- 1 1 methylcarbamate (Propoxur) in the gas phase. Computed with planar ( ππ*) and S1-pyramidalized ( nπ*) geometries. There are 1 CASPT2//CASSCF. two triplet states near the S1 state at the ππ* minimum, but they both have 3ππ* character and thus intersystem crossing to them should be negligible according to El-Sayed rules.54, 55 Similarly, 1 there is a triplet state near the S1 state at the nπ* minimum, but it has 3nπ* character and intersystem crossing to it will be negligible As discussed, the three-state model predicts that populating the for the same reason. Therefore, in the case of PA, the 1n* state is a requisite to reach the dissociative 1* state. photodissociation proceeds via S1, explaining the experimental Therefore, we may conclude that in molecules b to d, PFR is results.36, 37 unfavorable in the gas phase. The relative energy between the two

S1 minima helps to understand why Propoxur (d) diluted in different 56, 57 PFR sensitivity to substituents organic solvents does not undergo PFR, while it does in water. As it can be seen in Figure 4, for all molecules, the 1* state has approximately the same energy, while the 1n*state is strongly Besides phenyl acetate, the three-state model for PFR was applied destabilized by changing the methyl by an amino group. Thus, while to three other molecules to demonstrate the utility of this model in organic solvents, Propoxur behaves essentially as in the gas phase, to rationalize this type of reaction. In these additional molecules with 1n* > 1* (PFR unfavorable), in water, the interaction with the (see Figure 4-top), the methyl group attached to carbonyl moiety water oxygen disrupts the O-N correlation between n electrons, 1 1 56, 57 was replaced by an amino group, NH2 (in b and c) and NHCH3 in d, causing * > n* (PFR favorable). which results in a carbamate group. In molecules c and d, an Summarizing, the energetic balance between the 1*, 1n*, and 1 electron donor group was attached to the ortho position (OCH3 σ* states is critical for PFR (Figure 1). Depending on the 1 and OCH(CH3)2, respectively). The largest of these molecules (d), substituents (or solvent) the * can be stabilized relative to the known as Propoxur (or commercially as Baygon), is an important 1n* state, which leads to an overall reduction of PFR yield and an pesticide, for which its photochemistry holds a major practical increase of luminescence yield. Another substituent (or solvent) interest as a key to determine its fate in the environment. stabilizing the 1n* relative to the 1πσ* would lead to an increase of

CASPT2//CASSCF results for the S1 minima show that changing PFR yield and reduction of radical pair production. Thus, the three- the methyl by an amino group inverts the energetic order of state model for PFR can be applied to engineering compounds aiming the planar (1ππ*) and pyramidal (1nπ*) minima (see Figure 4). While at maximizing specific products and to rationalize experimentally with the methyl group, the 1nπ* state is more stable than the 1ππ* observed outputs. by 0.1 eV, with the amino group, the 1ππ* becomes the most stable by 0.2

Please do not adjust margins Please do not adjust margins

Conclusions 6. T. Höfler, T. Grießer, X. Gstrein, G. Trimmel, G. Jakopic and W. Kern, Polymer, 2007, 48, 1930. Our theoretical analysis provides a clear and comprehensive picture 7. T. Höfler, T. Grießer, M. Gruber, G. Jakopic, G. Trimmel and W. for PFR. We have shown that the bond cleavage is due to an Kern, Macromol. Chem. Phys., 2008, 209, 488. interplay of three singlet electronic states: an aromatic 1ππ*, 8. Y. Ishida, Y. Takeda and A. Kameyama, React. Funct. Polym., which absorbs the radiation and it is initially populated; a 2016, 107, 20. carbonyl 1nπ*, which transfers the electronic energy from the 9. M. Martignac, E. Oliveros, M. T. Maurette, C. Claparols and F. aromatic ring to the dissociative region; and a 1σπ*, Benoit-Marquie, Photochem. Photobiol. Sci., 2013, 12, 527. responsible for the homolytic cleavage. For phenyl acetate, the 10. A. C. Weedon and D. F. Wong, J. Photochem. Photobiol. A, 1991, 61, 27. transfers between these three states occur with a small barrier 11. A. Sanjuan, G. Aguirre, M. Alvaro, H. Garcia and J. C. Scaiano, between 1ππ* and 1nπ*, and without any significant barrier Appl. Catal. B, 2000, 25, 257. between 1nπ* and 1σπ*. Direct transfer from 1 1 ππ* to σπ* is 12. M. Passananti, M. Lavorgna, M. R. Iesce, M. DellaGreca, E. precluded by large energy barriers. Triplet dissociation is Criscuolo, A. Parrella, M. Isidori and F. Temussi, Env. Sci. Process. also not possible due to El-Sayed rules. After transferring to Impact, 2014, 16, 823. 1σπ* and reaching a conical intersection, the molecule may 13. C. K. Remucal, Env. Sci. Process. Impact, 2014, 16, 628. return to the parent species, dissociate, or form 14. Z. Wang, in Comprehensive Organic Name Reactions and cyclohexadienone intermediates, which are precursors for PFR. Reagents, John Wiley & Sons, Inc., 2010. In the gas phase and in nonpolar aprotic solvents, the ortho- 15. D. Iguchi, R. Erra-Balsells and S. M. Bonesi, Photochem. Photobiol. substituted product is obtained after slow hydrogen tunneling, Sci., 2016, 15, 105. while in protic solvents, intermolecular H shift should dominate. 16. J. C. Anderson and C. B. Reese, Proc. Chem. Soc., 1960, 217. 17. M. R. Sandner and D. J. Trecker, J. Am. Chem. Soc., 1967, 89, The three-state model for PFR provides a general picture 5725. beyond the PA prototype, as the photodissociation process in 18. M. R. Sandner, E. Hedaya and D. J. Trecker, J. Am. Chem. Soc., phenyl acetate should be analogous to that in other aromatic 1968, 90, 7249. 58-60 esters, amides, carbamates, and carbonates. Different 19. G. M. Coppinger and E. R. Bell, J. Phys. Chem., 1966, 70, 3479. substituents and solvents will naturally change the relative 20. J. W. Meyer and G. S. Hammond, J. Am. Chem. Soc., 1970, 92, energies of the 1ππ*, 1nπ*, and 1σπ* states, quantitatively 2187. altering the basic topography illustrated in Figure 1, and leading to 21. C. E. Kalmus and D. M. Hercules, Tetrahedron Lett., 1972, 1575. different rates and product yields. We have demonstrated, for 22. C. E. Kalmus and D. M. Hercules, J. Am. Chem. Soc., 1974, 96, 449. instance, that in the case of carbamate derivatives, PFR is 23. J. W. Meyer and G. S. Hammond, J. Am. Chem. Soc., 1972, 94. unfavorable in the gas phase (and likely in nonpolar aprotic 24. H. J. Yoon, S. H. Ko, M. K. Ko and W. K. Chae, Bull. Korean Chem. solvents too), because the predissociative state (1nπ*) is too Soc., 2000, 21, 901. 25. H. Kobsa, J. Org. Chem., 1962, 27, 2293. high in energy. Thus, these three states will always ultimately 26. D. Bellus, P. Hrdlovic and P. Slama, Collect. Czech. Chem. control absorption, energy transfer, and dissociation steps in PFR. Commun., 1968, 33, 2646. 27. H. J. Hageman, Tetrahedron, 1969, 25, 6015. 28. Schwetli.K, A. Mehlhorn, R. Noack and J. Stumpe, Zeitschrift Fur Acknowledgements Chemie, 1974, 14, 116. The authors thank CNPq and CAPES (PDSE program, 29. A. Mehlhorn, B. Schwenzer and K. Schwetlick, Tetrahedron, 1977, 88881.131599/2016-01) for scholarships and the support 33, 1483. from CESUP/UFRGS and Santos Dumont/LNCC, all funded by the 30. M. S. Syamala, B. N. Rao and V. Ramamurthy, Tetrahedron, 1988, 44, 7234. Brazilian government. The authors also thank the support of the 31. L. S. Kaanumalle, C. L. D. Gibb, B. C. Gibb and V. Ramamurthy, A*MIDEX grant (ANR-11-IDEX-0001-02) and the project Org. Biomol. Chem., 2007, 5, 236. Equip@Meso (ANR-10-EQPX-29-01), both funded by the 32. R. Suau, G. Torres and M. Valpuesta, Tetrahedron Lett., 1995, 36, French Government “Investissements d’Avenir”program. This work 1311. was performed using HPC resources from GENCI-CINES (Grant 2017- 33. A. K. Zarkadis, V. Georgakilas, G. P. Perdikomatis, A. Trifonov, G. A0010810012). G. Gurzadyan, S. Skoulika and M. G. Siskos, Photochem. Photobiol. Sci., 2005, 4, 469. 34. M. Gohdo and M. Wakasa, Chem. Lett., 2010, 39, 106. Notes1. P. Magnus and referencesand C. Lescop, Tetrahedron Lett., 2001, 42, 7193. 35. A. Zanutta, L. Colella, C. Bertarelli and A. Bianco, Optical 2. W. Bowers, T. Ohta, J. Cleere and P. Marsella, Science, 1976, 193, Materials, 2013, 35, 2283. 542. 36. S. J. Harris, D. Murdock, M. P. Grubb, G. M. Greetham, I. P. Clark, 3. F. Galindo, M. C. Jiménez and M. A. Miranda, in Arene chemistry: M. Towrie and M. N. R. Ashfold, Chem. Sci., 2014, 5, 707. Reaction mechanism and methods for aromatic compounds, ed. 37. S. Lochbrunner, M. Zissler, J. Piel, E. Riedle, A. Spiegel and T. Bach, J. Mortier, John Wiley & Sons, 2016. J. Chem. Phys., 2004, 120, 11634. 4. T. Magauer, H. J. Martin and J. Mulzer, Angew. Chem. Int. Ed., 38. S. Grimme, Chem. Phys., 1992, 163, 313. 2009, 48, 6032. 39. N. P. Gritsan, Y. P. Tsentalovich, A. V. Yurkovskaya and R. Z. 5. A. Rivaton, B. Mailhot, J. Soulestin, H. Varghese and J. L. Gardette, Sagdeev, J. Phys. Chem., 1996, 100, 4448. Polym. Degrad. Stab., 2002, 75, 17. 40. I. F. Molokov, Y. P. Tsentalovich, A. V. Yurkovskaya and R. Z. Sagdeev, J. Photochem. Photobiol. A, 1997, 110, 159.

Please do not adjust margins Please do not adjust margins

41. M. Gohdo, T. Takamasu and M. Wakasa, Phys. Chem. Chem. Phys., 2011, 13, 755. 42. T. Arai, S. Tobita and H. Shizuka, J. Am. Chem. Soc., 1995, 117, 3968. 43. T. Arai, S. Tobita and H. Shizuka, Chem. Phy. Lett., 1994, 223, 521. 44. D. A. Plank, Tetrahedron Lett., 1968, 9, 5423. 45. J. W. Meyer and G. S. Hammond, J. Am. Chem. Soc., 1972, 94, 2219. 46. J. Finley, P.-Å. Malmqvist, B. O. Roos and L. Serrano-Andrés, Chem. Phys. Lett., 1998, 288, 299. 47. G. Ghigo, B. O. Roos and P.-A. Malmqvist, Chem. Phys. Lett., 2004, 396, 142. 48. N. Forsberg and P.-Å. Malmqvist, Chem. Phys. Lett., 1997, 274, 196. 49. P.-O. Widmark, P.-Å. Malmqvist and B. O. Roos, Theor. Chim. Acta, 1990, 77, 291. 50. G. Zechmann and M. Barbatti, Int. J. Quantum Chem., 2008, 108, 1266. 51. F. Aquilante, J. Autschbach, R. K. Carlson, L. F. Chibotaru, M. G. Delcey, L. De Vico, I. Fdez. Galván, N. Ferré, L. M. Frutos, L. Gagliardi, M. Garavelli, A. Giussani, C. E. Hoyer, G. Li Manni, H. Lischka, D. Ma, P. Å. Malmqvist, T. Müller, A. Nenov, M. Olivucci, T. B. Pedersen, D. Peng, F. Plasser, B. Pritchard, M. Reiher, I. Rivalta, I. Schapiro, J. Segarra-Martí, M. Stenrup, D. G. Truhlar, L. Ungur, A. Valentini, S. Vancoillie, V. Veryazov, V. P. Vysotskiy, O. Weingart, F. Zapata and R. Lindh, J. Comput. Chem., 2016, 37, 506. 52. Z. Lan, W. Domcke, V. Vallet, A. L. Sobolewski and S. Mahapatra, J. Chem. Phys., 2005, 122, 224315. 53. R. Crespo-Otero, A. Mardykov, E. Sanchez-Garcia, W. Sander and M. Barbatti, Phys. Chem. Chem. Phys., 2014, 16, 18877. 54. N. J. Turro, Modern molecular photochemistry, Univ Science Books, 1991. 55. M. A. El-Sayed, J. Chem. Phys., 1963, 38, 2834. 56. G. A. Ana Sanjuán, Mercedes Álvaro, Hermenegildo García, J. C. Scaiano, Appl. Catal. B, 2000, 25, 257. 57. W. Schwack and G. Kopf, Z. Lebensm. Unters. Forsch., 1992, 195, 250. 58. M. A. Miranda and F. Galindo, in Molecular and supramolecular photochemistry: Photochemistry of organic molecules in isotropic and anisotropic media, eds. V. Ramamurthy and K. S. Schanke, Marcel Dekker Inc., New York, 2003, vol. 9, pp. 43. 59. D. J. Trecker, R. S. Foote and C. L. Osborn, Chem. Comm., 1968, 1034. 60. E. A. Caress and I. Rosenberg, J. Org. Chem., 1971, 37, 3160.

Please do not adjust margins